02/10 - Swifty Sunday!

8 to 1, yea his system is as good as always

 

Mine was revoked. I was told I was doing a great job and
- Ranking the real ending as the best is the main thing that I'm looking for. You ranked the real endings possible 80% of the time, unlikely 18% of the time, and gibberish 1%.

It was revoked and told me to do the qual test again if I want

 

That's possible, I am still getting all of this set up There's also some ambiguity because the AI this time around does a lot better at being mostly convincing

Happy to investigate more to figure out what a fair cutoff is. unfortunately, some spot checking is required - on this batch I had one worker guess randomly on ~3000, and it got autoapproved :/ so I had to increase the AA time for the future batches

 

No problem I sent the email that said I was doing a good job to you via PM. If I'm not giving you what your looking for no worries I understand, but if you could take a look when you get the time, I'd appreciate it.

 

block them and report them to amazon if it's possible - because that's really crappy Those types of people should not be working on the platform. Save them from scamming others.

 

would u mind checking mine too? i got dropped from 8 to 1, and I haven't done anything different since the previous fairly 'good' feedback

 

Title: Werewolves and Villagers Game (Bonus up to $1.00) [ $ | $ ] PANDA
Requester: Pivithuru [A2U0GLJ3CNX872]
TurkerView: [ $14.08 / hour ]
Description: You will participate in two "Werewolves and Villagers" games in this task. For each game, your base payment will be $0.25. Your rewards will be added and penalties will be deducted from the base payment. The minimum guaranteed payment for a game will be $0.05 and maximum possible payment will be $0.55. *This HIT should only be completed on a laptop or desktop computer. Do not accept if you are on a mobile device or tablet*.
Duration: 00:10:00
Available: 3
Reward: $0.10
Qualifications:

• HIT approval rate (%) GreaterThanOrEqualTo 95,
• Werewolves and Villagers Game Qualification EqualTo 1, [ Request | How to Use ]

[3VTCTNDILABMOSZV785NC0SN215TGZ]​

 

Greetings from Amazon Mechanical Turk,

The following is a summary of activity for your Mechanical Turk account for the week ending Feb 09, 2019.

Your HIT activity for this week:
- Number of HITs submitted: 239

Approvals and payments that occurred this week:
- Number of HITs approved: 228
- Number of HITs rejected: 0
- Bonus reward earned: $1.70
- HIT reward earned: $29.54
- Total Amount earned this week: $31.24

MTurk powers just haven't a clue that sometimes the "exception" to the "majority rule" is more important. ...

PLEASE KEEP OUR WEEKLY REPORTS !!!

 

Who knows what a fork is?

Enormous kitchen batch just went up

I don't qualify....

Perch Mturk

 

@rowan Here is my issue with the "you're picking the AI answer too often" logic

Spoiler

If all of that isn't an AI vomit of words, I'm shocked.

 

so more info that might clarify this:

Essentially I'm trying to get a bunch of AI-generated caption endings that humans know are unreasonable, but the AI thinks are fine. to do this, I'm starting with a bunch of endings that the AI thinks are good, and removing AI generated endings that are too good (and thus could be confused with real endings).

However, in order for me to trust your judgments, then you need to mostly prefer human written endings. If, say, you pick the real ending as 'best' 1/6th of the time, then you're likely just guessing wildly, and it doesn't mean as much if you say an ending is the best one

Anyways, hope that helps- if you have any ideas about how to make this clearer, lmk!

 

The following is a summary of activity for your Mechanical Turk account for the week ending Feb 09, 2019.

Your HIT activity for this week:
- Number of HITs submitted: 1,234

Approvals and payments that occurred this week:
- Number of HITs approved: 1,208
- Number of HITs rejected: 1
- Bonus reward earned: $20.00
- HIT reward earned: $244.85
- Total Amount earned this week: $264.85

How many of you full time Mturk? I'm coming back full steam after quitting my job and I'm wondering how much you guys spend each day, I feel I'm already at the computer most of the day anyway but I want to set a limit. right now I usually wake up at 10am est and try to go to at least 6pm but it's hard to tell once things start trickling off where Mturk ends and computer activities begin.

 

I think the whole "majority - unhelpful" narrative is a lie, they just don't want to fool with it even though it's completely automated.

 

A. Got caught cheating with the woman, extremely bitter about upcoming divorce and is following a scorched earth policy.

B. Alternatively, the rough draft script for "Lawnmower Man"

 

Some of the ones that I feel are a bit confusing are the ones that are like "The guy is skiing down a hill"

and some of the options are - "the ending credits are shown."

These are a bit tricky to me, as in context of the video - it's really impossible for me to say it's not true - especially if we don't have access to the video. So I always assume that these are true. But often times there will be multiple endings talking about the ending credits - and it's really tricky to figure out which one is human generated.

 

Sure, msg me your worker id

 

Josh Siegel [AATN5Y4RG8H0U] Academic survey on employee decision-making (~5-7 minutes) - $0.70 | PANDA

Low

Unrated

Unrated

$5.81 / hour
00:07:14 / completion time

Pros:

Cons:
Little longer than expected. When you check a bubble, the color is very pale which makes it look like you didn't check it.

Qualifications: Completed Pretest Already DoesNotExist ; Total approved HITs > 1000; HIT approval rate (%) > 98; Location EqualTo US;

To read Josh Siegel's full profile check out TurkerView!​